Research On Faults Diagnosis System Based On Hidden Markov Model

This paper researched on rotation machinery faults diagnosis system based Hidden Markov Models (HMM) .HMM dynamic pattern recognition theories and methods are studied, then researched on the system's function and realization methods and validate this system through experiment device. The details are studied as follow:Chapter one introduced the research meaning of faults detection and diagnosis of rotating machine. The technology of fault detection and diagnosis are introduced. The article sets forth the meaning of faults diagnosis system base on HMM. At last, the main contents and general structurescheme are presented.Chapter two introduced the basic ideas of Markov Chain theories and Hidden Markov Models (HMM) .Then the theories and algorithms of HMM are studied aid the application of speech recognisition uses HMM. At last, attempt to use HMM to faults diagnosis.Chapter three researched on base function and form of fault diagnosis system and research on the methods of realization and technology of fault diagnosis system based on HMM.Chapter four the methods of faults diagnosis, FFT-HMM and Wavelet-HMM, are studied based on the theory of HMM. Then researched the Discrete Hidden Markov Models (DHMM)and Continue Hidden Markov Models (CHMM) as different random functions. At last the experiment of run-up process of rotor machine is made to verify the effect of the faults diagnosis methods.Chapter five, the final part of this paper, summarizes the work of dissertation and presents conclusion and prospect.